Supplementary Materials and Methods for Enhanced air pollution via aerosol-boundary layer feedback in China T. Petäjä1*, L. Järvi1, V.-M. Kerminen1, A. Ding2*, J. Sun2, W. Nie1,2, J. Kujansuu1, A. Virkkula2,3, X. Yang2, C. Fu2, S. Zilitinkevich1,3,4 and M. Kulmala1* 1 Department of Physics, University of Helsinki, Finland 2 Institute for Climate and Global Change Research & School of Atmospheric Sciences, Nanjing University, China 3 Finnish Meteorological Institute, Helsinki, Finland 4 University of Nizhny Novgorod, Russia *Correspondence to: tuukka.petaja@helsinki.fi, dingaj@nju.edu.cn, markku.kulmala@helsinki.fi A1 Theoretical derivation of boundary layer height in polluted urban convective boundary layer A1.1 Energy balance The energy balance at the Earth surface reads (in W m-2): πΉ ∗ + πΉπ΄ = πΉπ» + πΉπΈ + πΉπ , (SI-1) where πΉ ∗ is the net radiation flux resulted from counteraction of solar and long-wave radiation fluxes, both of which consists of the upward and downward components; πΉπ΄ is positive (upward) energy flux due to anthropogenic heating of urban infrastructures; πΉπ» and πΉπΈ are turbulent fluxes of sensible and latent heat; and πΉπ is the net-storage heat flux. New advection is assumed to be negligible when compared to the other components. The difference ΔπΉ ∗ between the values of the net radiation flux in the non-polluted and polluted atmosphere leaves the anthropogenic heating unchanged, and causes variations βπΉπ» , βπΉπΈ and βπΉπ in all fluxes on the right hand side of Eq. (SI-1): βπΉ ∗ = βπΉπ» + βπΉπΈ + βπΉπ ≈ βπΉπ» + βπΉπΈ . (SI-2) Here, the variation βπΉπ in the net-storage heat flux is neglected as small compared to the major variations: βπΉπ» and/or βπΉπΈ ; and the sum βπΉπ» + βπΉπΈ represents the total turbulent heat-flux decrement caused by the air pollution. Clearly, the net radiation flux approaching the surface is smaller in the polluted atmosphere, so that βπΉ ∗ < 0. The major driving forth of the growth of convective BL is the near-surface value of the vertical π turbulent flux of the buoyancy b defined as π = π π + 0.61ππ, where g is the acceleration due to 0 gravity, π0 is the reference value of absolute temperature, θ is the potential temperature, and q is specific humidity. The buoyancy flux is expressed through the heat fluxes πΉπ» and πΉπΈ : πΉπ = π π π ππ0 (πΉπ» + 0.61 ππ π0 πΏπΈ πΉπΈ ) ≈ π π π ππ0 (πΉπ» + 0.07πΉπΈ ), (SI-3) where π is the air density, ππ is the heat capacity at constant pressure, and πΏπΈ is the latent heat of condensation. As follows from (SI-2) and (SI-3), to a rough approximation the decrement in the buoyancy flux βπΉπ caused by the air pollution can be taken to be proportional to the easily measurable βπΉ ∗ . A1.2 Characteristic height of urban convective BL as dependent on the surface buoyancy flux A typical well-mixed urban convective BL grows from the city border along the wind, according to the prognostic equation: πβ πΉ π ππ₯ = π π2πβ , (SI-4) where h is the BL height, x is the distance from the border, U is horizontal wind speed at the BL upper boundary, a is a dimensionless coefficient to be determined experimentally, and π = √ππ/ππ§ is the Brunt-Väisälä frequency in the free troposphere (37). If no relevant data is available, N can be taken equal to its typical climatological value for given city and given month. A reasonable rough estimate is N ≈ 10-2 s-1. Μ Μ Μ Μ Μ Μ is defined as the Reynolds-averaged product of The vertical turbulent flux of buoyancy πΉπ = π′π€′ fluctuations of vertical wind velocity (π€′) and buoyancy (π′). In the CBL, πΉπ is positive, which corresponds to the upward energy transport. In our analyses, this flux is determined directly from Μ Μ Μ Μ Μ Μ ; and it is measurements of turbulent fluxes of potential temperature Μ Μ Μ Μ Μ Μ π€′π′ and specific humidity π€′π′ demonstrated that πΉπ essentially reduces in the polluted atmosphere. Integrating Eq. (SI-4) yields: πΉ π₯ πΉ π· β = √2π ππ2 π ≈ √2π ππ2 π , (SI-5) where D is the horizontal city scale. The approximate relation (employing D as a characteristic distance) serves as a characteristic urban BL height. According to Eq. (SI-5), the ratio of the heights of the polluted and non-polluted urban convective BLs is equal to the square root of the ratio of the corresponding surface buoyancy fluxes: βππππ βπππππππ = √πΉ πΉπ,ππππ π,πππππππ . (SI-6) In windy weather, U is the speed of the background wind driven by the synoptic-scale pressure gradient. In calm weather, when the background wind is weak, the dominant role turns to the cityscale convective circulation cell driven by the urban heat island. Such cell produces its own convective wind estimated as π~ π΄(πΉπ β)1/3, (SI-7) where the combination (πΉπ β)1/3 is the Deardorff convective velocity scale, and A ~ D/h is the urban-cell aspect ratio that is essentially larger than unity. Then, substituting Eq. (SI-5) for U into Eq. (SI-6), yields the following alternative characteristic urban BL height: 2/3 β ≈ (2π πΉπ π· π΄π 2 3/7 ) , (SI-8) 2/7 which implies a weaker than square root dependence: h ~ πΉπ . A2 Correlation analysis Table 1 presents the correlation analysis for the radiation and flux parameters during the 8 month measurement period in Nanjing. The data presents cases when ambient RH is below 50% where the effect of hygroscopic water uptake of the aerosol particles remains small (36). Supplementary Table 1. Linear fits and associated correlation coefficients (R) between a) the aerosol mass (PM2.5) and b) BC concentration against the different radiation and flux components (y = a0x + a1). N is the number of data points. a) PM2.5 RH<100% RH<80% RH<60% b) BC RH<100% RH<80% RH<60% a0 a1 N R Kdown/Ktop Ldown/Ktop Q*/Ktop Fb/Ktop Kdown/Ktop Ldown/Ktop Q*/Ktop Fb/Ktop Kdown/Ktop Ldown/Ktop Q*/Ktop Fb/Ktop -3.315ο10-4 2.231ο10-4 -4.929ο10-4 -1.594ο10-7 -3.049ο10-4 1.429ο10-4 -5.237ο10-4 -1.813ο10-7 -6.144ο10-4 1.474ο10-4 -7.441ο10-4 -2.087ο10-7 0.537 0.348 0.385 4.889ο10-5 0.5705 0.3529 0.410 5.22ο10-5 0.653 0.350 0.460 5.699ο10-5 826 826 826 767 750 750 750 695 463 463 463 432 -0.07 0.22 -0.15 -0.34 -0.07 0.13 -0.17 -0.37 -0.16 0.11 -0.27 -0.36 Kdown/Ktop Ldown/Ktop Q*/Ktop Fb/Ktop Kdown/Ktop Ldown/Ktop Q*/Ktop Fb/Ktop Kdown/Ktop Ldown/Ktop Q*/Ktop Fb/Ktop -9.074ο10-3 3.942ο10-3 -11.495ο10-3 -3.164ο10-6 -8.268ο10-3 2.981ο10-3 -11.553ο10-3 -3.295ο10-6 -12.348ο10-3 2.118ο10-3 -15.709ο10-3 -3.962ο10-6 0.538 0.350 0.385 4.84ο10-5 0.573 0.353 0.410 5.10ο10-5 0.654 0.351 0.461 5.63ο10-5 837 837 837 777 754 754 754 699 460 460 460 429 -0.09 0.19 -0.17 -0.33 -0.09 0.14 -0.19 -0.35 -0.17 0.08 -0.29 -0.35 Fig. A1. Change in normalized a) solar radiation, b) longwave radiation, c) net all wavelength radiation and d) turbulent vertical flux of virtual temperature as a function of observed PM2.5 concentration. Ktop is the solar radiation at the top of the atmosphere (W m-2). Only daytime conditions from the non-rainy periods between 10:00 and 14:00 local time are considered. The colour of the data points indicate observed relative humidity. Fig. A2. A change in normalized a) solar radiation, b) longwave radiation, c) net all wavelength radiation and d) turbulent vertical flux of virtual temperature as a function of observed BC concentration. Ktop is the solar radiation at the top of the atmosphere (W m-2). Only daytime conditions from the non-rainy periods between 10:00 and 14:00 local time are considered. The colour of the data points indicate observed relative humidity.